Retuning Synaptic Dynamics Restores Memory Signals in Alzheimer’s Model

Alzheimer’s disease progressively strips away neurons and disrupts the synapses that encode memory, but a new computational study suggests that the surviving connections might be coaxed back into functional patterns by adjusting the timing rules that govern their short-term behavior. The research, published in Frontiers in Computational Neuroscience, is entirely a modeling study: no experiments in patients, animals, or laboratory tissue were performed.

Han Zhang, Yusuf Mohideen, and Chi Chung Alan Fung, of the City University of Hong Kong and the CityU Shenzhen Research Institute, built a spiking neural network using the NEST Simulator. The network contained 1,000 Hodgkin-Huxley excitatory neurons and 250 leaky integrate-and-fire inhibitory neurons, with embedded memory engram assemblies representing stored memories. Alzheimer’s pathology was modeled through accelerated neuronal loss and pathological hyperexcitability of the remaining cells.

The crucial ingredient was the Tsodyks-Markram model of short-term synaptic plasticity, which describes how synaptic strength changes transiently on millisecond-to-second timescales through facilitation and depression. Rather than a fixed property, the team treated the model’s two key time constants as adjustable dials: the facilitation time constant τfac, which controls how quickly synapses become temporarily stronger with repeated use, and the recovery time constant τrec, which governs how quickly depleted synaptic resources are replenished.

By sweeping these two parameters across a two-dimensional space, the researchers asked whether re-tuning them could rescue memory function in the damaged network. Engram recall fidelity was quantified by a selectivity ratio, Γ, and network-level oscillations were assessed from the power spectral density of the simulated local field potential, or LFP.

The results were striking in one direction. Increasing τfac from 500 milliseconds to 1,250 milliseconds rescued engram selectivity in the Alzheimer’s model at intermediate synaptic-weight regimes that were inaccessible at baseline settings.

The recovery time constant told a different story. The sweep revealed a clear trade-off: increasing τrec reduced the selectivity ratio Γ, whereas decreasing it maximized recall, with only small accompanying changes in background excitability. In other words, short-term facilitation and short-term depression modulation exert mechanistically distinct compensatory effects rather than interchangeable ones.

The authors frame these findings as identifying bounded parameter windows within which dynamic re-tuning of surviving synaptic circuits can meaningfully mitigate Alzheimer’s-related functional deficits, without requiring structural repair of the damaged network. The appeal of the approach lies in its target: short-term synaptic plasticity is a potentially targetable mechanism that operates on fast timescales, so compensating for circuit-level deficits dynamically might not depend on regrowing lost neurons or synapses.

Important caveats apply. The network is a simplified simulation of 1,250 neurons, and Alzheimer’s pathology was represented only as accelerated neuronal loss and hyperexcitability, not as the full molecular pathology of the disease. The study demonstrates what is possible within this computational framework; it does not establish that any intervention could achieve these parameter shifts in a human brain, nor does it identify a therapy. The authors themselves note that the claims expressed are solely those of the study’s authors.

Subject of Research: Computational modeling of short-term synaptic plasticity as a compensatory mechanism in Alzheimer's disease

Article Title: Short-term synaptic plasticity modulates the outcome of neurodegenerative diseases

Article References: Zhang, H., Mohideen, Y., & Fung, C. C. A. (2026). Short-term synaptic plasticity modulates the outcome of neurodegenerative diseases. Frontiers in Computational Neuroscience, 20, Article 1923602. https://doi.org/10.3389/fncom.2026.1923602

Image Credits: AI Generated

DOI: 10.3389/fncom.2026.1923602

Keywords: Alzheimer's disease, short-term synaptic plasticity, Tsodyks-Markram model, spiking neural networks, memory engrams, local field potential, NEST Simulator, neural network modeling, neurodegenerative diseases, synaptic facilitation, computational neuroscience, Short-term